Binary Image Enhancement Using Pai Model For Noise Suppression In Spatial Domain
I. Suneetha · i-manager’s Journal on Image Processing · 2014
Let f(x,y) be a digital image of size MxN with pixel values in the range [0, L-1], fn(x,y) be its noisy image corrupted by salt and pepper noise and g(x,y) be its denoisy image in which noise is suppressed. Noisy pixels in fn(x,y) can take the values of either minimum value or maximum value in the given dynamic range. Spatial filtering operation involves moving the centre of a square mask from pixel to pixel over the entire image. (1) In nonlinear filters, enhanced image g(x,y) at (x,y) is not linearly related to pixels in the neighborhood of input image. For each neighborhood, find pixel value based on the type of nonlinear filter and then map with the central pixel. Generally the size of the window is odd. The two basic filters are Maximum and Minimum filters. Max filter locates brightest point in an image so that it removes pepper noise only where as Min filter locates darkest point in an image so that it removes salt noise only as shown in Figure 1.